Morning Edition · Wednesday, September 30, 2026Published at 2:00 AM EDT · New York
Every private transfer is capped below $2,500 and co-signed by a separate protocol's sealed servers while the circuit fix waits for the V6 release.

Aztec Labs brought back zk.money on Monday, three years after shutting it down, as a self-custodial wallet for private stablecoin payments on the Aztec layer-2 network. Transfers between users hide amounts and counterparties from the public chain, while deposits arriving from Ethereum stay visible, CoinDesk reported. Each transfer must stay below $2,500, and a separate protocol called Oxide runs sealed servers that co-sign every send and withdrawal, according to The Block.
Those limits exist because the underlying network is running with a known defect. Aztec core contributors disclosed a critical vulnerability in the Alpha V5 proving system, found through internal artificial-intelligence-assisted auditing. An attacker could potentially construct a proof that passes verification for a transaction the network should reject, allowing the chain to accept a state transition outside the rules Alpha V5 is supposed to enforce. Contributors also state that they cannot determine from chain history whether anyone used the flawed path, because the system does not record enough information to distinguish ordinary accepted transactions from transactions accepted through the flaw. The fix is deferred to the V6 release, which is meant to include circuit updates that reject proofs tied to the affected system.
This is the failure mode that ordinary rollup safety mechanisms do not catch. Optimistic rollups rely on re-execution and fraud proofs. Validity rollups rely on the soundness of the circuit itself. When the circuit is unsound, that guarantee no longer holds, so the operator must substitute administrative controls, including value caps, application-level guards, and a co-signing service. Aztec is a testnet-stage alpha network and says so, but the design of zk.money shows what that trade-off means in practice. A user gets cryptographic privacy from the public, and for now, a non-cryptographic dependency on an external signer for both liveness and safety.
For context on scale, the layer-2 sector Aztec is entering secures $51.09 billion in total value according to L2Beat, with Base Chain at $16.19 billion and the largest zero-knowledge network in that set, Starknet, at $455.7 million. Aztec is betting that networks need confidential execution more than additional throughput. This disclosure shows what that bet costs while the proving stack is still being built.
Part of a tracked trend
Proving-System Bugs Become a Distinct Rollup Risk
Soundness defects in zero-knowledge proving systems will keep surfacing as a risk class separate from smart-contract exploits, because validator re-execution — the fallback most rollups rely on — cannot catch them, forcing teams into embargoed disclosure timed to upgrades and pushing users toward proof-system diversity and escape hatches.
Start a discussion in Townsquare.
More from this edition
Aztec gains a shipping narrative for its privacy layer 2 while conceding the caps, and rival privacy networks gain a contrast case for proof-system diversity.
The vulnerability disclosure is genuine and self-published, but it dates from 7 August (found 27 July), not this week, and the load-bearing detail is that Oxide's enclave co-signing means a theft would require a soundness fault and a compromised enclave at the same time, alongside a shared $50,000 daily deposit allowance the piece omits.
An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.
What this means
The mechanism is trust substitution. When a validity proof stops bounding state transitions, the guarantee shifts from mathematics to whoever operates the manual controls, in this case Aztec's transfer caps and Oxide's sealed co-signing servers. Users of zk.money, application teams building on Alpha V5, and any bridge exposed to Alpha V5 state carry that risk directly. Aztec faced a similar problem before, with the Alpha V4 vulnerability that forced users to withdraw ahead of a deadline in June, and the same pattern repeated at the next version. The likelier outcome over the coming months is that privacy and zero-knowledge networks ship with explicitly disclosed non-cryptographic backstops, caps, and manual override mechanisms as standard practice, and that competitors advertise proof-system diversity as a feature. What would prove that read wrong is a V6 release with independently reviewed circuit changes that lets Aztec remove both the value cap and the external co-signer without an incident.
What to watch
Observations to monitor, not financial advice.
Synthesized from: CoinDesk · Aztec Network (disclosure) · Aztec Network (Alpha V5) · The Block
Comments
2Sep 30, 6:00 AM · edited
Mandatory Oxide co signing on every send makes those sealed servers a liveness and censorship dependency, so the test at V6 is whether the $2,500 cap rises and the co signer requirement is dropped together.
Sep 30, 2:00 PM · edited
Capping transfers below $2,500 and routing them through another protocol's sealed servers means the co signer, not the circuit, is the actual security control, and it caps the anonymity set that makes the product worth using.